In a recent post, I noted that theories of Latin influence on Bezae’s Greek text were the norm before J. S. Semler argued that Bezae’s Greek had not been influenced by a Latin version (1767). Following Semler, both J. D. Michaelis and H. Marsh seem especially eager to disprove J. J. Wettstein’s hypothesis that D had been corrected to a Latin version.
J. D. Michaelis assembles readings that in his view “rescue the copyist from the charge of having corrupted the Greek from the Latin,” noting two categories of such readings, namely, those in which the Greek and Latin columns differ and those in which the Latin column has been altered from the Greek.  It is clear though that Michaelis’ evidence can only address the limited case in which the copyist of Bezae or its exemplar has been influenced by the Latin column during transcription.
For a case in which the Greek and Latin columns differ, Michaelis cites Mark 11:12, where D was at one time the only Greek witness known to have the singular participle (note that Legg lists Γ with + αυτον):
εξελθοντα απο βηθανιας
Yet d has the plural:
cum exissent (d with a i)
But while it is clear that d could not have influenced D, Michaelis points out that D has support for its singular participle in the Old Latin version, namely from Codex Veronensis:
cum exisset (b with c ff2 q r1)
So far from excluding the possibility that D might have been influenced by the Old Latin versions, Michaelis has unwittingly added support for Wettstein’s hypothesis! This reading suggests that, agreements aside, D has contact with a variant in the Latin tradition that is otherwise largely unknown in the Greek tradition.
Then for a case in which the Latin column has been altered from the Greek, Michaelis cites Acts 10:6, noting that the Latin version “in general” adds hic dicet tibi quid te oporteat facere, while d and e follow their Greek columns by omitting this phrase.
But there are some problems with the evidence.
First, the extra phrase is far from the general Latin reading. The other two Old Latin MSS extant in Acts 10:6 gig and p* both agree with d and e as well as much of the Vulgate tradition, while the addition is supported by pc and another part of the Vulgate tradition.
Second, Michaelis (laboring before Kipling’s edition of Bezae) observes that “the spurious addition ετος λαλησει σοι τι σε δει ποιειν is rejected from the Greek text of those two manuscripts [D and E].” But Bezae preserves here only the Latin side of Acts 10:6, the leaf is missing from the facing Greek. So although the point is moot anyway, we cannot actually know what D read. Yet Michaelis’ confidently-affirmed evidence continues to keep alive the notion that large-scale Latin influence is impossible in D. 
We might sum up Michaelis’ contribution in two points:
First, he seems to have influenced the way the question of Latinization was conceived, that is, primarily in transcriptional terms, while leaving unaddressed the larger possibility of systematic correction to a Latin version. In fact, transcriptional data cannot inform us on this question because it reaches no farther back than D’s exemplar. Nor can so-called “Latinisms” help, because in most cases there would have been no trace but the Latin reading itself.
Second, Michaelis and his successors seem to have assumed that he decisively answered Wettstein. But as we have seen, he does not address the same problem.
Clearly then Michaelis did not disprove the possibility that Bezae’s Greek text might have been influenced by a Latin version. The question must be regarded as still open. We simply cannot assume that Bezae’s most distinctive elements represent a pure Greek tradition.
 J. D. Michaelis, Introduction to the New Testament (trans. H. Marsh; 1802) 2/1:230. These examples are cited by D. C. Parker, Codex Bezae: An early Christian manuscript and its text (Cambridge: Cambridge University Press, 1992) 185.
 Parker follows Michaelis, observing “Ac 10.6, where d e follow D E in omitting the phrase hic dicet tibi quid te oportet facere which is found in the rest of the Latin tradition,” concluding that Michaelis “showed that the theory of Latinization was not able to solve the problem of the D text” (p. 185). But as we have seen, D is lacunose in Acts 10:6 and E d e are supported by the Old Latin and much of the Vulgate.